Bonding Optics to Metals

Structural adhesives for bonding optics to metals: a study of opto-mechanical stability

John G. Daly*, Damien J. Daly*Vector Engineering Inc.

ABSTRACT

With so many new adhesives available, characteristics affecting performance are not always well-defined. The user oftenselects an adhesive based on a single property and later finds his application compromised. This is an effort to studyrelevant properties ofseveral different structural4ype adhesives. The bonding geometry will utilize three types of glass(fused silica, BK-7, and Pyrex) bonded to metal mounts. The mounting geometry will include five different designapproaches. These designs will investigate: face bonding, counter-bored mounts, edge bonding, and a flexure mount. Thethree metals selected (aluminum, titanium, and invar) are not only common to the industry but often used for matching theCoefficient ofExpansion to the optical glass. Each optical fiat will have its reflective surface used as a reference forangular stability. The adhesives selected wilt compare more traditional epoxies with one-part Ultraviolet Light (UV) curedproducts. The obvious advantage ofthe UV-cured adhesives is the instant cure on-demand. Several adhesives have beenselected for differing properties including: viscosity, cure temperature, CTE, modulus ofelasticity, out-gassing, andshrinkage upon cure. Discussion will compare each adhesive, its properties, and ease ofuse. Angular stability will bemonitored as a function of: pre vs. post cure, accelerated life testing, thermal exposure, and vibration/shock exposure.Some discussion will be included on the Wavefront Distortion and Stress Birefringence.

Keywords: Adhesives, Bonding, Optics, Optical Stability, Structural Adhesives, Optical Mounts

1. INTRODUCTION

In the experience ofthese authors, very little has been published on the application ofadhesives for fastening opticalelements. Yoder1 has briefly addressed this subject in an excellent opto-mechanical text. Optical bonding has been acommon practice, but until recently extreme alignment stability requirements have been limited to a small portion of theelectro-opticat I photonics marketplace. Thousands ofadhesives are now available and the selection process is not alwaysclear. Priorities must be assigned to the properties in the determination ofa compromised-process. In order to providecontinuity and to aid the reader in understanding our adhesive selection evaluation, a short review oftbe critical adhesiveproperties is included.

Less than twenty years ago, the majority ofall optical bonding utilized a handftul ofadhesives. In general these weretypically 2 part epoxies with good adhesion, high strengths, and some resiliency. For example, 3M's Scotchweld 2216B/A 2 is still a favorite adhesive candidate. In addition to complying with these common properties, 2216's viscosity madefor easy application and its low outgassing is characterized as NASA space qualified. Some ofthe principle limitations ofthese 2-part adhesives included: critical mix ratio dependency; long room temperature cure or high temperature cures; arelatively high CTE (coefficient ofthermal expansion); and a high modulus ofelasticity. The high CTE and moduluscontributed to rigid bonds and high stress levels.

Users today expect much more from adhesives for fastening optical elements. They demand: low costs; excellentadhesion; instant cure; high strength; no stress; precise alignment stability; a wide temperature range; no out-gassing;and long life. Unfortunately, no single adhesive can meet all these requirements. It is the users responsibility todetermine where compromises can be made. With thousands ofUV-cure polymers available, the appeal ofinstant on-demand cure merits an evaluation of their properties. We have compared similar bonding geometries utilizing epoxies andUV-cure adhesives.

* contact: vectorengaol.com ; phone 1 352 383 5319; fax 1 352 383 3069: Vector Engineering mc, P0 Box 1287,Mount Dora, Fl 32776.

Optomechanical Design and Engineering 2001, Alson E. Hatheway, Editor,Proceedings of SPIE Vol. 4444 (2001) © 2001 SPIE · 0277-786X/01/$15.00 177

1.1 ADHESIVE HISTORY

The older established adhesives were developed for differing applications. Many ofthe traditional 2-part epoxy favored byoptical engineers were originally formulated and marketed for bonding either wood or metal. Adhesion and strength werethe principle properties the chemists pursued in these formulations. And in many cases, the adhesion was anticipated to beacceptable under conditions ofhigh temperature cure and high pressure loads. These attributes to the bonding process arenot consistent with most optical assembly processes.

Other adhesives for glass-to-glass bonding were independently developed. Early uses oforganics for cementing doubletswas succeeded by synthetic and more reliable adhesives. an3 developed a complete line ofpolymer-based adhesiveswith excellent optical properties that cured with UV (ultraviolet) radiation. This was well-suited to precise alignment IpositioningI centering followed by an instant cwe-on.demand process by exposure to UV radiation. Attempts to use theseadhesives for glass-metal bonding was met with limited success until more specific formulae were developed. Today,Norland, Dymax ', and many other vendors have UV-cure adhesives that rival the mechanical properties of2-part epoxies.

1.2 ADHESIVE TYPES

Adhesive types or categories do not follow any common nomenclature. However, most titles are descriptiveenough toavoid confusion. Glass-to-Glass adhesives imply transmissive optical quality but can often work for glass to metalsituations. In general, all adhesives are now synthetic. Common categories often utilize these terms: epoxy; polyurethane;acrylic; cyanoacrylate; silicones, etc. Other most application-orientated terms are: 2-part; i-part; RTV; UV-cure,..etc.

A structural adhesive definition in general has a tensile (or lapshear) strength that exceeds 1000 psi. This arbitrarystrength assignment is violated when some flexible RTV adhesives with lapshear strengths near 500 psi are used to secureoptics. Many times the exact formulae ofan adhesive are treated as proprietary, but the MSDS information can aid inpartial identification.

We have studied several 2-part epoxies in this study. Each will be listed with relevant properties as well as the processused in our tests. Several UV-cure polymers will be compared to the performance ofthese traditional epoxies.

1.3 ADHESWE PROPERTIES

This briefreview ofrelevant properties will aid the reader in understanding their influence on performance and how toevaluate candidates for selection.

Viscosity is defined as the resistance to flow or shear stress. This property is important for uncured adhesives. Theviscosity will influence its application and its ability to wet the surface ofthe substrates. High viscosity adhesives are easyto handle, control bead size and position. However, low viscosity improves the wetting or contact with the substrates.When the viscosity is too low, the adhesive will flow and may corrupt either the bond process or the clear aperture.Viscosity is directly proportional to temperature. Therefore caution is necessary when high temperature curie is employedand a correctly applied adhesive may flow and destroy the process.

Wetting is defined as the ability ofthe uncured adhesive to make contact with the substrate. This is critical for adhesion.Surface tension is often associated with this property. The adhesive must have a lower surface tension than the substrate forproper wetting to avoid beading. Pressure is also used to improve wetting and therefore adhesion.

After the adhesive has cured, its properties must be understood to determine compatibility with not only the materials butthe use and exposure ofthe assembly. We have focused on the properties that are most relevant to optical alignmentsensitive applications: durometer, strength, CTh, modulus, shrinkage on cure, the glass transition temperature, and out-gassing.

Durometer is a measure of the hardness where a relative scale of penetration is employed with a dedicated instrument. Acommon denominator for plastics, this can provide an indication of compliance or hardness of the cured adhesive.

Proc. SPIE Vol. 4444178

Strength is critical to performance and is typically measuredas: tensile; compressive; !apshear; peel; or cleavage. Ingeneral, lapshear tests are most common and provide adequate information. However, peel and cleavage can be commonmodes offailure when these loads are experienced Because ofthe high stress concentration in peel and cleavage loads,these strengths are much lower than tensile oflapshear. The bonding design should attempt to achieve loads free from peelor cleavage situations.

The CTE values become important for applications where exposure to wide temperature ranges exist. Ifpossible, the usertries to match the substrate CTE's first, and then use an adhesive with a value close to these. Differential expansion (orcontraction) will create optical element movement and also high stresses that can result in fracture ofthe optical element.The modulus ofelasticity plays a similar role in the selection ofan adhesive. Tensile moduli are normally available,however bulk and compressive can be determined. A high modulus adhesive that experiences either shrinkage on cure ordifferential expansion will experience proportionally high stresses.

AU adhesives experience some shrinkage during cure. Typical epoxy values are 3 to 5 %. This contributes to movementfor critical alignment applications. It also creates high stress levels. Suppliers are working on improved products withlower shrinkage values. We have studied some UV —cured adhesives with shrinkage on cure levels ofless than 0.2%

Polymer chains have limited mobility at temperatures below their Glass Transition Temperature. Above the Tg, wholechains are mobile and visco.elastic behavior results in diminished mechanical properties. The glass transition temperatureshould be factored into the selection ofany adhesive. The mechanical properties ofstrength, modulus, and Cit aretemperature dependent. Most suppliers will quote the CTE value for two temperature ranges: below Tg and above Tg. It isthe degradation in these parameters that is important and not the Tg, so the user must evaluate accordingly.

Outgassing is the release ofvolatile solvents. This occurs during cure but also throughout the life ofany material. NASAhas tested thousands ofmaterials and publishes a website with this information. The two measures ofFerformance are theTML ( Total Mass Loss, in %) and CVCM (Collected Volatile Condensable Material, in %). NASA has determined thatSpace-Qualified materials should have a TML < 1.0% and a CVCM < 0.1%. We have found these guidelines to be validand in application ofhigh fluence laser exposure additional testing may be conducted. Laser induced optical damage fromthe CVCM can be further evaluated by out-gassing and subsequent laser exposure tests .

2. METHODOLOGY

The adhesives selected were representative oftraditional epoxies with proven performance. In addition, several new UVcure adhesives were selected for comparison in similar applications. Each candidate adhesive is listed below with itproperties as well as the supplier source.

For simplicity, only one optical element geometry was tested. One inch diameter optical flats with a 0.25 inch thicknesswere used for most tests. Some thinner samples were available. The optical materials included BK..7, Pyrex, and fusedsilica. This is representative ofcommon materials with a wide range ofmechanical properties.

The metal substrates were defined in five geometries: (1) a flat plate; (2) three raised inplane pads; (3) a counterboredrecessed cell with a through hole; (4) a counter-bored opening with a through hole and side holes for injected edgebonding; (5) and a flexure designed mount. Three materials have been used for each geometry: black anodized aluminum(606l-T6); titanium (6AL-4AV) ; and gold-plated invar (invar-36). Figure 2. 1 — 2.3 show these fixtures. These solidmetal fixtures were designed to accommodate 4 optical test pieces independently (except the injection hole design with itsthree-optic geometry). In addition, at the center of each fixture the metal was machined with a V2 inch raised surfacewhich was diamond-polished to a mirror surface. This surface was the integral reference mirror from which angularmovement of the test optics could be compared. This design eliminated any dependency of the reference mirror stability.

Proc. SPIE Vol. 4444 179

Figure 2.3 Test Fixture #5 Flexure Mount

Figure 2.1 Test fixtures # 1 & 2: Flat Plates

Figure 2.2 Test Fixtures #3 & 4; Counterbore recessed


2.1 ANGULAR STABILITY TESTING

In many cases, the test optic was aligned and cured with its angular alignment measured postcure. Further temperatureexposure allowed alignment stability to be monitored at high and low temperature as well as after ambient equilibrium.The long term angular stability was monitored. The angular alignment was measured as the parallelism ofthe test opticssurfitce to the reference mirror surthee on the test fixture. For these tests we used a Nikon 6D Autocollimator7. Theprojected reticle provided a comparison ofthe reflected returns with a resolution ofbetter than 20 microradians. Newerinstruments with CCD arrays and computer interfiice are available but our work was limited to the older traditionalautocollimator method. In eases ofsevere distortion and optical element bowing, the image blur would limit resolutionand these conditions were noted in our data. It is not possible to quantify the surface deformation from this infonnation,but a subjective assessment was possible.

Data was collected in two formats. The optical element's first surface angular alignment was recorded with-respect4o thediamond polished reference mirror. The stability or retention ofthis initial alignment was measured under ambientconditions: after bonding, after thermal testing, and after storage times Thermal testing involved active monitoring of thealignment during temperature changes and at each stabilized temperature extreme. Typical data started at ambient,followed by a measurement at +55°C. Cold exposure at —30°C was recorded unless distortion limited resolution. Tn thatcase, data was collected at 0 °C.

3. DATA

Adhesive 2216 B/Agray

724-14C OP-61

Supplier 3M Ablestik Dymax

Type 2-part epoxy 2-parturethane 1-partComments High strength epoxy

and low outgassingFlexible high strength Multi-purpose glass to

metal bondingLow shrinkage and lowout-gassing

Cure Room temp Room temp Uv light

Viscosity 100,000 cps High 160,000 cpsWork Life 90 minutes 30 minutes nlaDurometer D 60 A 92 D 85

Shrinkage uponcure

3% (est.) 3% (est.) 0.3 %

LapshearStrengthAl-Al @ 25°C

2500 psi 1900 psi 2800 psi

Peel strength 25 piw N/a nlaGlass TransitionTemperature

40 0 C N/a 70 ° C

CTEBelow TgAbove T

.

102 x lO/°C134 x lO/°C

N/a 43x lO/°C59 x lO/OC

Modulus 2,400,000 psiTML 1.01 1.11 1.22

CVCM 0.05 0.12 0.02

TemperatureRange

-55°C to +100°C -55°C to +125°C -45°C to +170°C


Table 3.1 Adhesive Property Matrix

Material Aluminum Titanium InvarAlloy

-6061-T6 6AL4V Invar-36

Surface finish Black anodized Not plated Base: electroless NickelMil-A8625 Gold per Mil-std-

Type II, class 2 45204BType 2, class 00

(.00005 mm)Density 0.098 lb/in3 0.161 lb/in3 0.291 lb/in3CTE 23.9x10/ °C 8.46x10/ °C 0.9x10/ °CYoung's Modulus

•

10,000,000 psi 16,500,000 psi 21,500,000 psiTensile Yield Strength 3 1 ,000 psi 130,000 psi 70,000 psi

Table 3.2 Metal Substrate Mechanical Property Matrix

Material Pyrex BK-7 Fused SilicaComments Inexpensive high Borosilicate glass with Synthetically fhsed

temperature glass excellent optical quality quartz for critical opticalrequirements

Density 2.23 g/cc 2.53 g/cc3 2.2 g/ccCTE 3.6xl0/°C 7.1x10/°C 0.56x10/°CYoung's Modulus N/a I 1,700,000 psi 10,600,000 psiTensile Yield Strength 3,000 psi 1,000 psi 7500 psiIndex ofrefraction, 1 .473 1 .5 168 1.459dOptical transmission 325 nm to 2500 nm 325nm to 2500 nm 190 nm to 4400 nm

Table 3.3 Optical Element Property Matrix

4. RESULTS

As summarized in our conclusions, the data presented in this paper represent only preliminary results. The followingtable shows initial trends from configuration changes in bonding a one-inch diameter flat optical element in an aluminummount.

In each case, 3 adhesive bond points were used. This common practice accomplishes the following: definition ofa plane;controlled and consistent adhesive application; minimum use ofadhesive; and adequate bond area for strength. The onlymount material tested has been aluminum. This black anodized surface was cleaned with acetone prior to bonding.

The three glass materials represent not only common materials, but a wide range ofCTE values. Ratios ofCTE for glass-to-Aluminum are: fused silica (42.7), Pyrex (6.6), and BK-7 (3.4). All adhesives were cured at room temperature.Therefore the CTE difference will manifest itself during thermal testing. The high temperature test at +55°C was lesssevere than the — 30 °C test since the temperature deltas were: 28 vs. 57 °C.

The adhesive used in these preliminary tests are defined above. Our objective to evaluate these newer instant cure UVadhesives with respect to the traditional epoxies can be assessed from the data summarized below. We found theperformance ofOP6l to match the epoxies. Since the mechanical properties ofthe UV-cured adhesive equal theconventional epoxies, their use can be recommended. Early tests with another UV-cure adhesive Dymax 0P30have started since the very low modulus offers low stress applications.


Glass Adhesive Long term ambient Angular movement CommentsType stability from —30°C to-I-55°C

Aluminum Flat PlateFusedSilica

3M 2216 B/A < 15 prad For 0°C to +55°C< 60 iradnodata@-30°C

Large CTh mismatch causeddistortion at —30°C.

FusedSilica

AbleBond724-14C

< 10 prad < 200 irad This more flexible adhesive reduceddistortion but lacked stability.

BK-7 3M 2216 B/A < 45 irad < 80 irad Closer CTh match improvedperformance.

3 RaisedPads on AluminumPyrex-

AbleBond724-14C

< 150 pirad < 85 prad Well defined geometry providesimprovement relative to flat plate.

Pyrex OP 61 < 100 jirad < 20 pradCounterbore Recess in Aluminum

Pyrex 3M 2216 B/A < 50 irad < 100 prad Similar performance. Mostmovement from more flexible 724-14C adhesive.Pyrex AbleBond

724-14C< 50 irad < 200 ptrad

Pyrex OP 6 1 < 70 irad < 80 iradBK-7 3M 2216 B/A < 30 prad < 50 irad Improvement from better CTE match

for BK-7 vs. Pyrex.BK-7 AbleBond724- 14C

< 50 prad < 200 irad

BK-7 OP 61 < 70 irad < 90 .tradInjection Side Holes in Counterbore Recess in Aluminum

BK-7 3M 2216 B/A < 15 prad < 80 irad Results similar to counterbore butmuch easier to bond and to controladhesive on edges only.

BK-7 AbleBond724-14C

< 50 pirad < 90 jrad

BK-7 OP 6 1 < 45 irad < 100 irad3 Pad Flexure Mount in Aluminum

BK-7 3M 2216 B/A < 80 pirad < 45 jirad All adhesives result in very stableperformance during temperatureexposure.

BK-7 AbleBond724-14C

< 60 prad < 60 prad

FusedSilica

AbleBond724- 14C

< 100 jirad < 50 irad

FusedSilica

OP-6l < 30 prad < 30 prad

Pyrex 0P6 1 < 80 irad < 60 prad

Table 4.1 Test Data and Results Summary

With strengths similar to epoxies, the alignment retention stability shows similar results justifjing theirconsideration for potential use. Adhesive shrinkage upon cure will also influence the alignment stability. Highshrinkage will create optical element movement during cure as well as high stress.

5. CONCLUSIONS

As this paper deadline approached, we continued to record data. Definition and setup for these tests coupled with sometime constraints relegated this paper to preliminary test results. However, this paper successfi.illy defines our objectivesand test methodology. We will continue to record data and refine our tests until a fill range ofresults for each optical


element and metal substrate is compared for epoxies and UV-cured adhesives. We also have scheduled tests with newexperimental UV cure adhesives that have combinations oflow CTE, low shrinkage on cure, high 1g. and high strengths.

The authors will present updated results at the SPIE annual meeting later next month. As well, the authors remain availablefor subsequent questions concerning these tests and our results.

From the preliminary results, the following statements can be made1) These new UV adhesives have similar lapshear, tensile, and peel strengths to epoxies.2) Several UV adhesives have much lower shrinkage on cure values.3) Several UV adhesives have lower CTh values.4) Many UV adhesives have low out.gassing and are NASA space-qualified.5) Our tests show that the optical element alignment retention for these UV adhesives is comparable to the

traditional epoxies.

From these preliminary results, the high strengths ofOP-61 and its instant cure make is a realistic candidate to replaceepoxies for optical mounting where alignment stability is required. Other UV-cured products with similar properties areexpected to perform well and our continued tests will address this subject.

The Flexure mount tested is similar to the work ofBacich8. The results for these tests clearly show that this design isextremely stable over thermal exposure. Similar angular stability was recorded for all three varying Cit value opticalmaterials. This type ofmount was easily thbricated with a final step ofEDM (electrical discharge machining) and can besuccessftilly employed where stable mirror alignment is required over a wide temperature range.

ACKNOWLEDGEMENTS

The authors must acknowledge the Dymax Corp. TorrIngton, CT. for flmding for this effort. They requested anindependent evaluation oftheir UV-cured adhesive products as substitutes for epoxies in opto-mechanical applications. Inaddition to funding and adhesive samples, Dymax also provided valuable technical support. Northrup Grumman LaserSystems, Apopka, FL provided access to their optical assembly equipment and environment test equipment. Intellitec, adivision ofAPD, Deland, FL. also supported this effort by providing access to their optical and environmental testequipment.

REFERENCES

1. Yoder, P.R, Opto-Mechanical Systems Design, 2nd ed., Marcel Dekker Inc. N.Y. 1993.2. 3M Center, Adhesives Div. Bldg. 2208E-05, St. Paul, MN 55144. www.mmm.com.3. Norland Products, 695 Joyce Khmer Av., New Brunswick, NJ 08902. www.norlandprod.com4. Dymax Corp., 51 Greenwoods Rd., Torrington, CT 06790. www.dymax.com5. NASA, Goddard Space Flight Center, Materials Engineering Branch. http://epims.gsfc.nasa.gov/og6. Hovis et. a!., Optical Damage at the ..., SPIE Vol. 2114, 145. 1993.7. Nikon USA, at 1 800 645 6687. www.nikon.com8. AbleStik, 20021 Susana Rd., Rancho Dominguez, CA 90221. www.ablestick.com9. Bacich, J. "Precision lens mounting" U.S. Patent No. 4,733,945, 1988.


Documents

Bonding Optics to Metals